Elon Musk’s xAI Launches Grok Code Fast 1 With Agentic Coding Abilities, Free For A Week

Elon Musk’s AI startup xAI has launched a new lightweight coding-focused model, Grok Code Fast 1, that comes with advanced agentic coding capabilities. Designed as a full-stack coding agent, it can handle both backend and frontend tasks. The company claims the model is faster and more responsive than rival coding tools. For the first week, starting August 28, developers can use it for free on selected platforms. Also Read: Elon Musk Reveals Big Gaming Plans: xAI To Launch ‘AI-Generated Game’ By 2026

Built on a New Architecture

According to xAI, existing coding models often slow down when performing reasoning loops or tool-calling tasks. Grok Code Fast 1 is built to address this issue by offering faster performance for agentic workflows. While the firm did not disclose the technical details of its architecture, it revealed that the model was trained using programming-related data for pre-training and refined with real-world pull requests during post-training. Also Read: Why Thousands Are Cancelling Netflix After Elon Musk’s Latest Tweet

The company also worked with several third-party platforms to make sure the model behaves reliably during real coding sessions. Interestingly, Grok Code Fast 1 can call tools such as grep, terminal, and file editing, giving developers more flexibility while working. Also Read: Elon Musk’s xAI Fires 500 Employees? Report Suggests Major Layoffs

Language Support and Performance

The model is versatile too. It works across the full software stack and supports popular languages like TypeScript, Python, Java, Rust, C++, and Go. According to xAI, it can finish common programming tasks with very little oversight. On benchmark tests, Grok Code Fast 1 managed a 70.8% score on the SWE-Bench-Verified benchmark, which measures coding ability.

Grok Code Fast 1 Pricing and Access

Until September 4, developers can try Grok Code Fast 1 without any cost through platforms such as GitHub Copilot, Cursor, Cline, Kilo Code, Roo Code, opencode, and Windsurf. The company has not confirmed if rate limits will apply during the trial period.

Once the free week ends, pricing will be set at $0.20 (approx. Rs 17) per million input tokens, $1.50 (approx. Rs 132) per million output tokens, and $0.02 (approx. Rs 1.7) for cached input tokens.